Thermoelectric Chalcostibite: Solid-State Synthesis and Thermal Properties

Article information

Abstract

1. INTRODUCTION

Thermoelectric devices directly convert heat into electricity and are suitable for sustainable energy recycling. Thermoelectric technology provides opportunities to address global energy demands and the depletion of fossil fuels [1]. The performance (efficiencies) of thermoelectric devices are primarily described by the dimensionless figure of merit (ZT) [2]:

where α is the Seebeck coefficient, ρ is electrical resistivity, κ is thermal conductivity, and T is absolute temperature. Based on the relationship, to increase ZT a high power factor (α²/ρ: high Seebeck coefficient and low electrical resistivity) and low thermal conductivity are required [3].

The thermal conductivity consists of the electronic (κ_e) and lattice (κ_ph) thermal conductivities. Electronic thermal conductivity is related to charge carrier transport, while the lattice thermal conductivity is affected by phonon scattering [4,5].

The thermal conductivity is expressed as [6]

where d is density, c_p is specific heat, and D is thermal diffusivity.

Bi₂Te₃- and PbTe-based thermoelectric materials are used in solid-state cooling devices and power generation systems, but expanding the market is difficult because they use expensive, rare, and toxic elements [7]. However, Cu-based ternary chalcogenide compounds have attracted considerable attention as thermoelectric materials because of their low thermal conductivity, low cost, and environmental friendliness [8]. Specifically, among Cu-SbS compounds, Cu₃SbS₄ (famatinite), Cu₃SbS₃ (skinnerite), Cu₁₂SbS₁₃ (tetrahedrite), and CuSbS₂ (chalcostibite) have been reported [9]. Chen [10] reported that isolated pairs of Sb 5s electrons (lone-pair electrons) generate local distortions in the crystal structure, and the differences in the bonding angles of Cu-Sb-S cause the lone-pair electrons to move away from the nucleus due to Coulomb repulsion. Chen also interpreted that the larger the S-Sb-S bonding angle, the lower the expected thermal conductivity. Cu₃SbS₃ has the largest bonding angle, exhibits the lowest thermal conductivity, and Cu₃SbS₄ has the highest thermal conductivity because no lone-pair electrons exist.

CuSbS₂ exhibits intrinsic p-type conduction characteristics because of the existence of Cu vacancy defects with a shallow ionization level and low formation energy. It also has a bandgap energy of 1.5 eV, which makes it a promising thermoelectric material at intermediate temperatures [11,12]. CuSbS₂ exhibits an orthorhombic structure in the Pnma space group [13]. Kamimizutaru et al. [14] reported that the ZT value of pure CuSbS₂ was as high as 0.5 at 700 K, which was thought to be due to the low thermal conductivity derived from the lone-pair electrons of Sb³⁺. In this study, the optimal process conditions for chalcostibite synthesis-consolidation using mechanical alloying (MA) and hot pressing (HP) were investigated, and the thermal properties were examined and compared with those of other Cu-Sb-S compounds.

2. EXPERIMENTAL PROCEDURE

MA was employed to synthesize chalcostibite via a solidstate route, where elemental powders of Cu (purity 99.9%, < 45 mm), Sb (purity 99.999%, < 150 mm), and S (purity 99.99%, < 75 mm) were used as the starting materials. The mixed powders and stainless-steel balls were placed in a surface-hardened stainless steel jar, and then after forming an Ar atmosphere inside the jar, MA was performed at a rotation speed of 350 rpm for 6–24 h using a planetary ball mill (Pulverisette5, Fritsch). The synthesized chalcostibite powder was loaded into a graphite mold and consolidated by HP at 573–673 K for 2 h under a compression pressure of 70 MPa in a vacuum.

Phase analysis of the MA powders and HP specimens was conducted by X-ray diffraction (XRD; D8-Advance, Bruker) using Cu Kα radiation, and diffraction peaks were measured over a 2θ range from 10° to 90° at intervals of 0.02°. The phase was identified and its lattice constant was calculated using Rietveld refinement (TOPAS program). The thermal stability and phase transition of the chalcostibites were examined using thermogravimetric–differential scanning calorimetry (TG–DSC; TGA/DSC1, Mettler Toledo). The microstructures of the sintered specimens were observed by scanning electron microscopy (SEM; Quanta400, FEI). The atomic distributions of the component elements were confirmed using energy-dispersive spectrometry (EDS; Quantax200, Bruker) by elemental line scanning and mapping. The thermal properties of the chalcostibites prepared under each MA–HP process condition were evaluated. The thermal diffusivity was measured using the laser flash method of TC-9000H (Advance Riko) in a vacuum. The thermal conductivity was evaluated using Equation 3 based on the measured thermal diffusivity. The density was measured and calculated by the Archimedes method, and the sintered density (relative density) was compared with the theoretical density of chalcostibite (4.93 g cm^-3) [10]. The specific heat of chalcostibite as reported in the literature (0.40 J g^-1 K^-1) [10] was used to ensure the reliability of the thermal conductivity evaluation.

3. RESULTS AND DISCUSSION

Figure 1 shows the XRD results for the CuSbS₂ powders fabricated using the MA process. A single chalcostibite phase was synthesized by MA at a rotation speed of 350 rpm for 6 h (MA350R6H), and no specific phase transitions were observed even when MA was conducted for 24 h (MA350R24H). All the samples exhibited an orthorhombic structure (space group: Pnma) consistent with the standard diffraction data (PDF# 01-073-3954).

Fig. 1.

XRD patterns of CuSbS₂ synthesized by MA.

As Table 1 shows, the lattice constants of the chalcostibite powder obtained by MA were a = 0.6018 nm, b = 0.3796 nm, and c = 1.4495 nm, which were similar to those reported by Kyono and Kimata of a = 0.6016 nm, b = 0.3797 nm, and c = 1.4499 nm [15]. The full width at half maximum (FWHM) of the diffraction peak for the (013) plane (2θ = 29.9°) of the MA350R12H specimen was as large as 0.45032°. This was attributed to particle refining and residual stress caused by MA.

Table 1.

Lattice constants, FWHM, and relative densities of CuSbS₂.

Figure 2(a) shows the TG analysis performed to investigate the thermal stability of the CuSbS₂ MA powders. Rapid mass loss occurred at temperatures of approximately 800 K or higher, due to the melting of CuSbS₂ and the volatilization of S. According to the TG analysis of chalcostibites by Esperto et al. [16], a mass loss was observed at temperatures above 824 K. Figure 2(b) shows the DSC analytical results of the CuSbS₂ MA powders according to the MA time. One endothermic reaction occurred at 814–815 K for all specimens, which corresponded to the melting point of chalcostibite. Hobbis et al. [17] also observed a mass reduction and an endothermic reaction at 800 K in the TG– DSC analysis of CuSbS₂. Pi et al. [18] reported that mass loss occurred rapidly above 645 K in the TG–DSC analysis of Cu₁₂Sb₄S₁₃ (tetrahedrite), and endothermic reactions were recorded at 853 K (decomposition of tetrahedrite) and 886 K (fusion of tetrahedrite). Lee and Kim [19] observed one endothermic peak at 873 K in the DSC analysis of Cu₃SbS₃ (skinnerite), which corresponded to the melting reaction of skinnerite. Lee et al. [20] reported endothermic reactions at 829–830 K and 880 K in the DSC analysis of Cu₃SbS₄ (famatinite), which corresponded to the melting temperatures of famatinite and skinnerite, respectively. Chen [10] reported that the melting points of CuSbS₂, Cu₃SbS₃, and Cu₃SbS₄ were 826, 880, and 910 K, respectively. In this study, we synthesized a thermally stable chalcostibite compound without any phase transitions below its melting point. The XRD (Figure 1) and TG–DSC (Figure 2) results showed that the optimal MA condition to synthesize a single CuSbS₂ phase was determined to be MA350R12H (350 rpm and 12 h).

Fig. 2.

(a) TG and (b) DSC analyses of CuSbS₂ MA powders with respect to MA time.

Figure 3 shows the XRD patterns of CuSbS₂ sintered by HP using the MA350R12H powder. All HP samples contained (retained) a single chalcostibite phase consistent with the MA powder, regardless of the HP temperature (573-673 K). Table 1 lists the lattice constants and FWHM values of the HP specimens. Compared with the MA powder, the lattice constants did not change: a = 0.6017–0.6018 nm, b = 0.3801 nm, and c = 1.4495–1.4497 nm. However, the FWHM₍₀₁₃₎ of the HP specimens was significantly reduced to 0.25972–0.36144°. This was due to the stress relief and grain growth caused by the HP process at high temperatures.

Fig. 3.

XRD patterns of CuSbS₂ sintered by HP.

Figure 4(a) shows the TG analytical results for the HP specimens. Mass loss occurred at approximately 800 K or higher but was less than that of the MA powder (Figure 2(a)). Although the mass loss derived from the melting of chalcostibite and the volatilization of S, it was relatively low because of the densely sintered compact. Figure 4(b) shows the DSC analysis of CuSbS₂ with respect to HP temperature. Only one endothermic peak was observed at 818–821 K for each specimen, which was interpreted as the melting point of chalcostibite. The melting point increased by 3–6 K as compared with that of the MA powder specimen. This was assumed to be caused by the slight compositional change derived from S volatilization during the HP process and stress relaxation by heat treatment. In this study, single-phase chalcostibite was synthesized and retained during the MA– HP process, and secondary phases and phase decompositions and transformations did not occur.

Fig. 4.

(a) TG and (b) DSC analyses of CuSbS₂ HP specimens with respect to HP temperature.

Figure 5 shows SEM images of the polished and fractured surfaces of the HP samples. Some pores were observed in the HP573K2H specimen. However, when the HP temperature increased, dense microstructures were obtained and crystal grains were grown. As Table 1 shows, high relative densities of 94.6–99.7% were obtained that compared with the theoretical density of chalcostibite (4.93 g cm^-3) [10].

Fig. 5.

Polished and fractured surfaces of CuSbS₂ HP specimens.

Figure 6 shows the results of the elemental analysis of the HP623K2H specimen hot-pressed at 623 K for 2 h. From the EDS elemental line scans and maps, no secondary phases were found besides the chalcostibite phase, and its component elements (Cu, Sb, and S) were uniformly distributed in the phase.

Fig. 6.

SEM-EDS micrographs with elemental line scans and maps for the HP623K2H specimen.

Figure 7 shows the thermal diffusivity of CuSbS₂ as a function of temperature. As the measurement temperature increased, the thermal diffusivity of chalcostibite decreased, and with the HP623K2H specimen, the thermal diffusivity was (0.75–0.36) × 10^-2 cm² s^-1 at 323–623 K. The thermal diffusivities of skinnerite (Cu₃SbS₃), famatinite (Cu₃SbS₄), and tetrahedrite (Cu₁₂Sb3S4) in the same temperature range were reported to be (0.29–0.34) × 10^-2 cm² s^-1 [19], (0.54– 0.29) × 10^-2 cm² s^-1 [20], and (0.32–0.37) × 10^-2 cm² s^-1, respectively [21]. For skinnerite and tetrahedrite, the thermal diffusivities increased slightly as the temperature increased, but the temperature dependence was low. However, the thermal diffusivities of chalcostibite and famatinite decreased rapidly with increasing temperature.

Fig. 7.

Temperature dependence of thermal diffusivity for CuSbS₂

Figure 8 shows the thermal conductivity of CuSbS₂ as a function of temperature. The thermal conductivity was evaluated from the data shown in Figure 7 and Table 2 using Equation 3. When the HP temperature increased from 573 K to 673 K, the thermal conductivity increased from 1.00 W m^-1 K^-1 to 1.68 W m^-1 K^-1 at 323 K. This was interpreted as an increase in thermal diffusivity due to grain growth and crystallinity improvement, even though the density and specific heat of the specimen were not changed by the HP process at high temperatures. However, the thermal conductivity decreased to 0.62-0.72 W m^-1 K^-1 at the measurement temperature of 623 K. Compared with the thermal diffusivity of other Cu-Sb-S compounds at 323–623 K (0.64-0.74 W m^-1 K^-1 for skinnerite [19], 0.73-0.83 W m^-1 K^-1 for tetrahedrite [20], and 1.15-0.63 W m^-1 K^-1 for famatinite [21]), the thermal conductivity of chalcostibite was similar to that of famatinite in value and temperature dependence.

Fig. 8.

Temperature dependence of thermal conductivity for CuSbS₂.

Table 2.

Densities and specific heat values of Cu–Sb–S ternary compounds for thermal conductivity evaluations.

When the temperature increased, the Cu-Sb-S compounds exhibited a decrease in thermal conductivity (T^-1 dependence) under the influence of the Umklapp process (phonon-phonon scattering) and had low values of less than 1 W m¹ K^-1 above 500 K [10,22]. The lone-pair electrons of Sb in the Cu-Sb-S compounds (except famatinite) had a significant effect on thermal conductivity and resulted in a difference in thermal conductivity in the low-temperature region. This was due to the lattice anharmonicity caused by the electrostatic repulsion between the lone-pair electrons and adjacent chalcogen ions (S^2-) [9]. In the case of famatinite, Cu₃SbS₄ exhibited a relatively high thermal conductivity because the electrons in the outer shell of the Sb atom were completely combined to form a tetrahedron with SbS4; thus, no lone-pair electrons were observed [8]. However, skinnerite with lone-pair electrons exhibited a large S-Sb-S bonding angle of 99.3°, and the Sb 5s lone-pair electrons resulted in very low thermal conductivity due to Coulomb repulsion [8,19]. The S-Sb-S bonding angles of tetrahedrite and chalcostibite were 95.7° and 95.8°, respectively, which were similar to each other but lower than that of skinnerite. However, tetrahedrite exhibited a lower thermal conductivity than skinnerite because of the additional phonon scatterings caused by the Cu atoms of the triangular CuS₃ bonds in Cu₁₂Sb₄S₁₃. Skinnerite and tetrahedrite had lower thermal conductivities than famatinite, and no significant change was observed in their thermal conductivity with increasing temperature [16,23].

4. CONCLUSIONS

Thermoelectric chalcostibite CuSbS₂ compounds were synthesized by MA and HP, and their phases, microstructures, and thermal properties were examined based on the process parameters. Through XRD phase identification and TG–DSC thermal analysis, the optimal conditions for the preparation of chalcostibites were determined to be MA at 350 rpm for 12 h and HP at 623 K for 2 h under 70 MPa. When the mechanically alloyed powder was hot-pressed, X-ray diffraction peaks sharpened because of stress relaxation and grain growth. The thermal diffusivity and conductivity increased as the HP temperature increased but decreased as the measurement temperature increased. Compared with the thermal properties of other Cu-Sb-S compounds (skinnerite, famatinite, and tetrahedrite), chalcostibite exhibited higher thermal diffusivity and conductivity in the low-temperature region. However, at high temperatures, the thermal conductivity decreased due to phonon scattering, reaching low values of less than 1 W m^-1 K^-1 above 500 K. Cu-Sb-S compounds exhibited low thermal conductivity values because of the presence of lone-pair electrons and large S-Sb-S bonding angles; chalcostibite also exhibited low thermal conductivity values at high temperatures. In this study, the orthorhombic chalcostibite phase was fabricated using the MA–HP method as a solidstate process and was confirmed to be a potential thermoelectric material.

Article information Continued